proceedings - irbnet.de · cib working commission w078 information technology in construction...
TRANSCRIPT
-
CIB W078 - Information Technologyfor Construction
CIB Publication 361
Proceedings
W078 - Special Track 18th CIB World Building Congress
May 2010 Salford, United Kingdom
-
CIB WORKING COMMISSION W078 ‐ INFORMATION TECHNOLOGY IN CONSTRUCTION
PAPERS AND POSTGRADUATE PAPERS FROM THE SPECIAL TRACK
HELD AT THE CIB WORLD BUILDING CONGRESS 2010, 10‐13 MAY 2010
THE LOWRY, SALFORD QUAYS, UNITED KINGDOM
Selected papers from the Proceedings of the 18th CIB World Building Congress. Proceedings edited by: Professor Peter Barrett, Professor Dilanthi Amaratunga, Dr. Richard
Haigh, Dr. Kaushal Keraminiyage and Dr. Chaminda Pathirage
W078 Special Track Papers (excluding Postgraduate Papers) reviewed by: Dr. Alain Zarli, Dr. Bill East, Mr. Ersen Firat, Dr. Dana Vanier, A/Prof. Jos van Leeuwen, Ms. Kathryn Davies, Dr.
Yum Kwok‐Keung, Dr. Matt Prins, Dr. Nicola Maiellaro, Dr. Marc Bourdeau, Dr. Marja Naaranoja, Prof. Zhiliang Ma, Mr. Praveer Kumar, Dr. Sami Kazi, Dr. Thomas Grisham and
A/Prof. Robert Amor
CIB Publication 361
-
W078 ‐ INFORMATION TECHNOLOGY IN CONSTRUCTION
PAPERS AND POSTGRADUATE PAPERS FROM THE SPECIAL TRACK The objectives of the Working Commission are to foster, encourage and promote research and development in the application of integrated IT throughout the life‐cycle of the design, construction and occupancy of buildings and related facilities, to pro‐actively encourage the use of IT in Construction through the demonstration of capabilities developed in collaborative research projects and to organise international cooperation in such activities and to promote the communication of these activities and their results. The aim of W078's work is broad in terms of the design, construction and occupation and occupancy of constructed facilities, but primarily it relates to the integration and communication of data, information and knowledge in the facility's life cycle.
-
CONTENTS Papers A Building information Modelling Based Production Control System for Construction 1 Sacks, R. Radosavljevic, M. Barak, R. RCM‐Plan: A Computer Prototype for Improving Planning Reliability from a Lean 14 Production Viewpoint Gonzalez, V. Alarcon, L.F. Ulloa, H. Building Information Modelling in the Netherlands; A Status Report 28 van Nederveen, S. Beheshti, R. Willems, P. Radio Frequency Identification (RFID) and the Lean Construction Process 41 Taylor, J.M. The Last Mile: Best Practices for Successful Implementation of Mobile Communication 54 Technologies at the Construction Site Lasker, G. Cox, R.F. Orczyk, J.J. Converse, D. Real‐Time Management in a BIM Model with RFID and Wireless Tags 67 Sattineni, A. Empirical Application of GPS Fleet Tracking Technology to a Soil Excavation Process 76 Kang, J. Ahn, S.M. A Synopsis of the Handbook of Research in Building Information Modeling 84 Isikdag, U. Underwood, J. A Comperative Aanalysis of the Strategic Role of ICT in UK and Turkish Construction 97 Industries Underwood, J. Isikdag, U. Goulding, J. Kuruoglu, M. Digital Image Processing for Evaluating Defect Level in Visual Quality Inspection 110 Peansupap, V. Lorfor,C. Towards a Smart, Energy‐Efficient ICT‐Empowered Built Environment: The REEB 121 Strategic Research Agenda Zarli, A. Bourdeau, M. Hannus, M. Hassan, T. Postgraduate Papers Building Information Modelling Processes: Benefits for Construction Industry 137 Olatunji, O.A. Sher, W.D. Gu,N. Ogunsemi, D.R.
-
Impact of Information Technology to Facilitating Communication and Collaboration 152 in Libyan Public Sector Organisations Bezweek, S. Egbu, C. Building Information Modelling: Literature Review on Model to Determine the Level of 168 Uptake by Organisation Haron, A.T. Marshell‐Ponting, A. Aouad, G. Agent‐Based Negotiation Mechanism for Automatic Procurement in Construction 185 Ho, C. Shih, H. Preliminary Performance Evaluation of an ORDB‐Based IFC Server and an RDB‐Based 192 IFC Server Using the Bucky Benchmark Method Jeong, J. Lee, G. Kang, H. A Conceptual Framework for Research in Spatial Data Sharing 201 Salleh, N. Khosrowshahi, F. The Changing Perception in the Artefacts used in the Design Practice through BIM 212 Adoption Coates, P. Arayici, Y. Koskela, L. Usher, C. CIB Brochure 224 Disclaimer 226
-
A Building Information Modelling Based Production Control System for Construction
Technion, S.R.
Israel Institute of Technology
(email: [email protected])
Radosavljevic, M.
University of Reading
(email: [email protected])
Technion, B.R.
Israel Institute of Technology
(email: [email protected])
Abstract
We propose visual interfaces for BIM-based construction management systems to empower
construction personnel on site by providing them with construction process and status information.
Computer-aided visualization, not only of the construction product, but also of the construction
process, can provide a unique service to support decision-making by workers, supervisors and
managers, with the goal of achieving stable flows. The three main user interfaces – for a) detailing of
work packages into finer-grained task definitions by the trade managers and preparation of proposed
weekly work plans, for b) collaborative planning and integration between the plans of the different
trade crews, and for c) day to day communication of product and process information to and from
work crews – have been designed and implemented as functional mock-ups. They have been
evaluated in three focus group workshops with project engineers, construction site supervisors, trade
crew leaders and logistics managers. The findings at this early stage are that the interfaces provide
rich information for production control, including monitoring of current process status, and fulfil the
guiding principles defined for BIM-enabled production control. However, significant R&D is still
needed for back-end integration of the various information system components of the system
architecture before the system can be implemented and tested on site.
Keywords: building information modelling, lean construction, production management, visualization
1
mailto:[email protected]
-
1. Introduction
With few notable exceptions, the majority of academic and industrial research on computer-aided
design and visualization in construction has dealt with building design and with pre-construction
planning. There has been far less effort to develop Building Information Modelling (BIM) based tools
to support coherent production management on site. The neglect of production management on the
part of researchers of IT in construction reflects the decline in attention paid to production
management on the part of general contractors and construction managers (Ballard 2000). For various
reasons, construction companies have adopted a business practice of reducing core staff to a
minimum and implementing work through subcontracting (Sacks and Harel 2006). At the same time,
lean construction thinking applied to construction production systems has increased awareness of the
benefits of stable work, of pull flow of teams and materials to reduce inventories of work in progress,
and of process transparency to all involved. 3D visualizations of process status and future direction,
delivered to all on site, are either essential or at least highly beneficial for all of these (Formoso,
Santos et al. 2002). They can empower people working on site to manage the day to day flow of
construction operations with less direct control from higher levels of management, with better quality
and less waste (Sacks, Treckmann et al. 2009).
Production control in construction on site can be facilitated through use of the Last Planner System™
(LPS) (Ballard 2000). In prefabrication projects, methods such as the Process Planning Methodology
(PPM) have proved effective (Radosavljevic and Horner 2007) . Application of the LPS enables trade
managers and construction engineers to collaboratively prepare weekly construction plans that are
feasible and have a reasonably high chance of being executed as planned. The system works by
empowering those who carry direct responsibility for executing work to participate in planning the
work. It is based on the principles of flow defined in lean production texts and the Transformation-
Flow-Value TFV theory (Koskela 2000) of production in construction.
However, the LPS does not overcome all of the difficulties nor does it remove all of the waste
inherent in construction. In practice, percent plan complete (PPC) measures do not reach 100%
(research has shown that the best sites achieve approximately 80% PPC (Bortolazza, Costa et al.
2005)). By definition, lean systems are always subject to continuous improvement (Womack and
Jones 2003), and the LPS is no exception. One of the reasons for this is that construction systems are
uncertain and subject to process change within the time frame of the weekly planning window, so that
filtering tasks for maturity on a weekly level cannot ensure complete process stability. Another is that
the delivery of product and process information to workers can at times be ineffective or inefficient:
product information is provided in the form of drawings and specifications, which contain
inaccuracies or errors. Process information is scant, inaccurate and incomplete: trade crews are
generally uninformed about delays in material deliveries, unavailability of equipment previously
committed to them, or changes in the work plans of crews working in their vicinity. Where they are
informed, it is often too late for them to adapt their own plans. Seppanen (Seppanen 2009) provided
empirical evidence of the systematic failure of traditional production control to manage short-term
decision-making concerning trade crews' progress through a building, resulting in unstable plans and
low productivity.
2
-
Sacks et al. (Sacks, Radosavljevic et al. 2010) proposed to address these shortcomings by increasing
the degree of resolution for planning and responsive re-planning to a daily level, with the support of a
production planning, control and feedback information system based on building information models.
The system proposed is called ‘KanBIM’, denoting the implementation of a lean production system
with pull flow control (symbolized by the Kanban method) (Ohno 1988) using a building information
modelling (BIM) (Eastman, Teicholz et al. 2008) based information system.
2. Background
A BIM-based lean production management system for construction must enable:
1. visualization of the construction process and its status;
2. visualization of the construction product and work methods;
3. support for planning, negotiation, commitment and status feedback;
4. implementation of pull flow control;
5. maintenance of work flow and plan stability;
6. formalization of production experiments for continuous process improvement.
These principles emphasize the role of a KanBIM system in supporting human decision making,
negotiation among trade crews to coordinate weekly work plans, reduction of the granularity of
planning to a daily level, real-time evaluation of task constraints to compute task maturity, and
implementation of the language/action perspective.
Some BIM solutions have expanded their capabilities by adding 4D functionality. Among these are
'Tekla Structures' (Tekla 2008) and ‘Virtual Construction’ (VICO 2007). Since Tekla's core
functionality is detailing steel and precast concrete structures, its construction management tool
emphasizes fabrication and delivery control based on planned erection dates. As with most 4D BIM
tools, users can link tasks to model objects and use critical path methods to schedule them. To
support fabrication and delivery control Tekla allows users to schedule each piece, within the task,
individually. Virtual Construction’ integrates a BIM model with Location Based Scheduling (LBS),
which uses an underlying CPM network solver. It is based on a bill of quantities and a set of working
rates and costs for resources that are linked to tasks, and enables representation of the schedule as a
line of balance chart as well as in Gantt chart form. These tools, and others like them, include 4D
model visualization but do not support the collaborative production level planning that is essential for
trade managers and crew leaders on site.
Specialized 4D construction planning software, such as 'Synchro Professional' (Synchro 2007),
provide project scheduling, construction visualization, synchronization with design changes, supply
chain management and virtual construction simulation. They do not have internal scheduling
capabilities or an integrated BIM solution but instead allow users to import both schedule and 3D
model from various other applications.
3
-
A small number of applications have been developed to support Last Planner System™
implementations, but they do not use building models to support visualization. CICLOPS (Evolution-
IP 2009) is an internet application that allows its users to collaboratively prepare and control weekly
work plans. CICLOPS calculates the percent plan complete (PPC) for each weekly plan and can
perform 'Non-Completion Analysis' based on the user's recording of the reasons for tasks being
delayed. ‘WorkPlan’ is a planning tool, developed in research (1999), that applies a database of work
packages and constraints to support work planning. SPS (Koerckel and Ballard 2005) is a commercial
package that helps reduce supply chain variations.
The LEWIS system (Sriprasert and Dawood 2003) represents the most advanced attempt to date to
compile a construction production management system that fulfils the KanBIM principles. However,
it falls short in making the process status visible to work teams on site, it does not explicitly facilitate
negotiation and collaboration in work planning, and it does not implement pull flow control of work.
Thus most 4D solutions incorporate scheduling tools and a connection to a BIM model, but their core
use is for planning and visualizing the process. With the exception of LEWIS, they are not intended
for ‘real time’ production control. They lack the ability to detail work plans with sufficiently fine-
grained resolution, and they have no tools for delivery of information to the work face or reporting
from it, or for assisting real-time decision-making during the course of production itself.
3. Research goal and method
The KanBIM concept encompasses a holistic approach to embedding lean production control
processes through delivery of both process and product information to all project participants,
specifically including workers at the work face on site, using a building information model as its
backbone. Since no comparable systems exist, and no existing software could be adapted for the
purposes of evaluation of the proposed KanBIM system, the research method involved three steps
that were performed in three iterations, with the system being refined and re-evaluated in each cycle:
a) Process analysis and system design;
b) Programming of functional mock-ups of its interfaces;
c) Evaluation of the system in focus group workshop evaluation sessions.
System design began with preparation of a detailed ‘future state’ process flow map of the work flow
envisaged for production planning and day to day production control on construction sites. The
information system required to support the process was then derived, and defined in a system
architecture plan. This step also required selection of the delivery methods (hardware) for each
interface.
Functional mock-ups were prepared for the three main user interfaces. The mock-ups sought to
provide sufficiently complete functionality to thoroughly demonstrate the system’s intended modes of
operation. These user-interfaces cover the stages of a) preparation by trades for weekly work
4
-
planning meetings, b) negotiation between trade crews prior to and during weekly work planning
meetings with the construction management team, and c) day to day interaction with trade crew
leaders on the job site. Programming of the functional mock-ups served not only the evaluation step,
but was in and of itself a formative activity in testing the assumptions made in defining the work flow
and the system architecture, applying to them a rigor that could not have been achieved otherwise.
The user interfaces were evaluated in three focus group workshops which each involved construction
managers, trade crew managers, and crew leaders, held in the UK and in Finland. The remainder of
this paper describes the first two aspects.
4. The KanBIM planning and control process
The process chart shown in Error! Reference source not found. describes the actors in construction
site production management, the information they each generate, a set of ‘activity scenarios’ in which
information is generated, and the way the information is distributed and recorded in the different
information repositories. The process starts with the creation of a Master Plan. In this stage the users
compile and maintain a set of high-level activities and subordinate work packages, and schedule
them, including trade assignments and buffering. High-level resource levelling must also be done for
major equipment and spaces. This is done using existing construction planning tools.
Stop Task
Prepare Look Ahead Plan
PrepareMaster Plan
21
8
Start
Master Planning
Look Ahead Planning(make ready process)
Weekly Work Planning(compile, coordinate/negotiate and
commit)
Daily Work(commit, execute and report)
Compile and Detail Tasks
3
Coordinate & Commit to Weekly
Work Plan
Start Task
5
Negotiate proposed
plan
Yes
Manage Logistics
Task in Progress
Report Task Completion
No
Construction Planner
Section/zone managers
Construction Planner
Section/zone managers
Trade managers
Trade manager
Trade crew leader
Construction Planner
Section/zone managers
Trade managersTrade crew
leadersHealth & Safety Logistics Manager
Construction Planner
Section/zone managers
Health & Safety
Trade crew leader
Trade crew leader
Logistics Manager
Trade crew leader
Site Engineers Inspector Section/zone
manager
End
Problemsduring
execution
No
Inspect Work and Task
Completion
10
Confirm completion
Yes
Figure 1: Process flow model for the KanBIM system, showing defined activities 1 to 10
5
-
The next stage is look ahead planning (see Error! Reference source not found.). It consists of
breaking down the high-level activities into smaller, manageable work packages, defining logistic and
engineering constraints in the form of connections between activities and assigning equipment and
materials. The master plan and the look ahead plan are done by managers of the general contractor (or
construction management company) and the principal work package subcontractor managers. Both of
these stages are the same as standard LPS, with only one additional requirement, which is that they
are prepared using a BIM interface in which building elements are associated with the activities. This
capability, available in the existing commercial software described above, allows integration of the
product model with the high-level process model. Since 4D functionality is increasingly common in
BIM tools, we assume that an integrated product and process model can be prepared and consider it
as the starting point for the next stages.
The next step of the LPS process, weekly work planning, is divided here into two stages. First, in
activity 3 in Error! Reference source not found., each trade crew details its work packages into a
set of candidate tasks that it can perform during the following week, in preparation for the weekly
work planning meeting (stage 4 in Error! Reference source not found.). This activity starts with a
set of candidate work packages that were drawn from the look-ahead plan according to their planned
start date and priority. Each work package contains a set of 'task types' representing the different
kinds of work needed to perform it according to the production method. For example, in order to
erect a drywall, the following tasks are needed: build the wall frame; close the first side with plaster
boards; place insulation materials and fix any mechanical, electrical or plumbing (MEP) embeds;
close the second side with boards; and apply joint strips, sand and paint. BIM objects can require one
or more task types and the associations are recorded with the object's properties.
The work packages are shown using symbols and highlighted object groups in the model. The trade
contractor’s manager and his or her crew leaders divide the work packages into candidate tasks by
selecting a subset of building elements from the work package elements and grouping them into
distinct tasks according to their task types. For easier selection and better control of the overall
process of dividing the work packages into tasks, all building elements that have not yet been
allocated to tasks are labelled 'unassigned' and highlighted appropriately. The user interface to
support this activity is shown in Figure 2, which shows a hierarchical work package/task type/task
tree, a view of the building model focused on the work package zone and elements with symbols
representing its tasks, and a weekly schedule planning area at the bottom of the screen. Tasks are
scheduled and assigned to crews by dragging their symbols to the rows of specific crews on specific
days. In addition to tasks created and assigned by the trade manager, there are also two kinds of
special tasks: tasks that the trade manager assigns to other supporting trades and tasks that are
assigned to this trade subcontractor by other trades. These tasks need to be assigned to crews in order
to become part of the weekly work plan, in a negotiated process that is explained below.
Since each trade contractor creates its own proposed weekly work plan, the plans need to be
synchronized and finalized to form a mutually agreed project-wide work plan. This is done in a
weekly work planning meeting (activity 4 in Error! Reference source not found.) that is directed by
the project planner and in which all trade managers participate. During the meeting the project
planner reviews the candidate work packages and tasks for promotion to approved tasks for the
6
-
coming week. The interface for this activity is presented on two large screens: a data view (shown in
Figure 3) and a corresponding model view. The two screens are merely different representations of
the same content (one alphanumerical and the other graphical) and any operation on one, is
automatically reflected in the other. For example, when a task is selected in the data view the model
view will focus on and highlight its building elements and show temporary equipment; or when the
tasks are filtered in one view (by date, space, contractor etc.) the other will show the same results.
The interface allows the users to switch between four different aggregation data views (tasks sorted
by contractors, work packages, spaces and shared equipment) to eliminate any clashes and to improve
plan reliability.
Figure 2: User interface for detailing work packages to tasks and compiling the weekly work plan by
allocating crews to tasks
Any conflicts identified must be resolved through discussion and coordination between the relevant
trade managers. To resolve conflicts they can change their proposed plans using the same interface
used for initial planning (Figure 2). Changes could include rescheduling tasks, assigning more crews
or workers, changing resources by changing construction methods, and others. The changes are made
while all the participants are online so that the project planner views and all 3D model views will
reflect the changed overall weekly work plan.
After applying changes to the plan to make it feasible and acceptable for all the ‘last planners’, each
of them must explicitly accept their part of the plan and commit to executing theirs tasks. Plan
acceptance is shown on the project planner interface and only when a group consensus is achieved is
the weekly plan approved as a whole.
7
-
The next level of planning takes place on a daily basis, concurrently with execution of the work
through each week. This is the heart of the KanBIM process, where the crew leaders are given direct
access to the work plan and empowered to coordinate their work with all other crews as and when
needed (activities 5, 6 and 8 in Figure 1). The specialized model interface (shown in Figure 4Error!
Reference source not found.), which shows each crew leader’s specific tasks, is delivered via a
large scale touch screen (see Figure 5). This interface not only delivers process and product
information on demand, it also collects process information in real-time. Crew leaders use it to report
the start of tasks as they are begun, to update ongoing tasks according to actual performance, to report
that they have stopped work on a task and report the problem that caused the stoppage, and to report
completion of finished tasks.
Figure 3: Project planner contractor view interface for creating integrated and synchronized weekly
work plans
Problems that adversely affect execution, such as unavailable equipment, can be reported together
with details that enable responders to resolve them, such as details of which specific piece of
equipment is malfunctioning or missing, as shown in Figure 6. In this way crew leaders can also
report design issues directly on the model by using graphic annotation tools and voice messages. The
production management server can alert those responsible for solving the issue according to a
predefined work flow and create action items for fixing it. In the event that a crew leader needs to
change the execution sequence of his/her tasks, they can use this screen to initiate dialog to negotiate
the changes with the project planner and any other relevant crew leaders, in order to maintain overall
plan stability. Any changes are immediately reflected in all model views, so that all project
participants are aware of actual current status.
8
-
For learning purposes and to improve project performance, when a task is reported complete crew
leaders are asked to report any difficulties even if the task was completed as planned. By pressing the
complete button, the crew leader is also pulling an inspector to approve the completion of the task
(activity 10 in Figure 1). If the task completion is rejected, the rework needs to be re-scheduled by the
project planner and the trade manager.
Figure 4: Trade crew leader work status and reporting interface showing a crew's tasks. The crew
leader can ask the system to show neighbouring tasks for a complete picture of the overall work
Figure 5: Work face specialized model
interface on a 40” touch-screen mounted on a
mobile trolley. The system identifies crew
leaders (by RFID reader or by entry of a
unique ID code) and delivers specifically
tailored information concerning their tasks.
Figure 6: Reporting form for problems during
execution which led to stopping a task. The
reporting tool enables information flow from
the work face to the information servers to
update the work status and to raise flags when
problems are encountered.
The information for each task is organized in a 'control card' according to seven pre-conditions and
constraints: preceding activities, workspace, information (designs and specifications), safety,
9
-
materials, equipment and crew. For each pre-condition an independent maturity index (MI) is
calculated based on the constraints release status, so that a user can 'see' the maturity status of any
given task. Full details are provided in (Sacks, Radosavljevic et al. 2010).
5. System architecture
Figure 7 provides a high-level view of the system architecture. The main database contains the
construction model, which is a combination of the product model, the process model and the status
model. At the start of any project, they are generated from the design and fabrication models by
applying construction methods (recipes), work package aggregation and compiling temporary
equipment process related objects. Subsequently, the construction BIM modeller is responsible for
synchronization of the database with the design and fabrication models. Interaction between the
KanBIM users and the construction model is facilitated by user interfaces such as look ahead
planning (based on 4D capabilities), weekly plan preparation (Figure 2), weekly work planning and
negotiation (Figure 3), crew leaders' interface for delivering information and reporting status (Error!
Reference source not found.) and an alert system to support organizational work flow. All of these
are based on lean construction processes.
Two separate modules work in the background. The first module generates tasks constraints as soon
as tasks are created. As most constraints are predefined at a higher level for work packages, this
module details the constraints at the task level. The second module computes the maturity index (MI)
and a pull flow index (PFI) for each task. The PFI defines the priority to be assigned to a candidate
task according to the need for that task as determined by the maturities of its successor tasks, which
reflect the downstream demand, or pull.
The sources of the information the system uses extend beyond the boundaries of the construction
product and process mode. Information may reside in different peripheral construction management
systems, such as logistics, purchasing, human resources and personnel control, design management
systems, fabrication management systems and external databases. Sophisticated information or
objects brokers are needed to integrate this information.
10
-
Figure 7: System architecture schema
6. Conclusions
The principles for development of a KanBIM system have been classified in seven main areas (Sacks,
Radosavljevic et al. 2010): process visualization; product and method visualization; computation and
display of work package and task maturity; support for planning, negotiation, commitment and status
feedback; implement pull flow control; maintain work flow and plan stability and formalize
experimentation for continuous improvement. Some of the ways in which the KanBIM system, as
specified to date, fulfils these principles, are discussed below.
During plan execution, current status visualization is attained using the set of graphical symbols
shown in Figure 4. The symbols describe the current task status: ready, not ready, task in progress,
task stopped, etc. Symbols that represent deviation from plan are supplemented with additional
information, such as maturity level or partial completion indicator.
The BIM is the foundation of the KanBIM system database. A 3D model view serves as a background
platform in all interfaces for conveying project data and navigating through it. The challenge is to
make product and process information ubiquitous at the workface without encumbering crew leaders
or workers with hardware that may hamper their comfort, safety or productivity. This can be achieved
using personal digital assistants, mobile phones or other portable wireless devices, but these all have
limitations, particularly with regard to screen size. The primary solution suggested for implementing
KanBIM interfaces is to use large format all-weather touch-screen monitors which do not impose
physical restrictions on workers, enable discussion among crews who can all view the same model or
animation together, and provide the essential function of easy-to-operate online feedback. This format
also enables easy navigation and data access.
11
-
The KanBIM system deals with plan stability on two levels: the planning process and the execution.
In planning, it uses the maturity index as the main parameter for deciding which work package or task
will be done during the week. In task detailing (stage 3 in Figure 1), a task can initially be assigned to
a weekly plan even if its maturity is not yet 100%, but this implies a commitment on the part of the
trade manager to release all constraints by the planned execution date. During execution, the KanBIM
system works to maintain plan stability by applying the principle of ‘sticking to plan’ while at the
same time enabling rapid negotiation and thorough coordination of any necessary changes to the plan.
The pitfalls of potential negative impacts on other trades and the danger of ‘making-do’ and
subsequent rework mean that plan changes must be negotiated and recorded. The system enables
negotiation by facilitating ad-hoc toolbox meetings within a crew with real-time information, or
conversations between all those who might be influenced from rescheduling the task so that the new
plan will not compromise their work.
References
Ballard, G. (2000). The Last Planner™ System of Production Control. School of Civil Engineering.
Birmingham, U.K., The University of Birmingham: 192.
Bortolazza, R. C., D. B. Costa, et al. (2005). A quantitative analysis of the implementation of the Last
Planner System in Brazil. 13th Conference of the International Group for Lean Construction. R.
Kenley. Sydney, Australia, UNSW: 413-420.
Choo, H. J., I. D. Tommelein, et al. (1999). "WorkPlan: Constraint-Based Database for Work
Package Scheduling." Journal of Construction Engineering and Management 125(3): 151-160.
Eastman, C. M., P. Teicholz, et al. (2008). BIM Handbook: A Guide to Building Information
Modeling for Owners, Managers, Architects, Engineers, Contractors, and Fabricators. Hoboken, NJ,
John Wiley and Sons.
Evolution-IP. (2009). "CiCLOPS." from http://www.ciclops.info/.
Formoso, C. T., A. D. Santos, et al. (2002). "An Exploratory Study On The Applicability Of Process
Transparency In Construction Sites." Journal of Construction Research 3(1): 35 - 54.
Koerckel, A. and G. Ballard (2005). Return on investment in construction innovation - a lean
construction case study. 13th Annual Conference on Lean Construction, Sydney, Australia, UNSW.
Koskela, L. (2000). An exploration towards a production theory and its application to construction.
VTT Building Technology. Espoo, Helsinki University of Technology: 296.
Ohno, T. (1988). Toyota production system: beyond large-scale production. Cambridge, Mass.,
Productivity Press.
12
http://www.ciclops.info/
-
Radosavljevic, M. and R. M. W. Horner (2007). "Process planning methodology: dynamic short-term
planning for off-site construction in Slovenia." Construction Management and Economics 25(2): 143-
156.
Sacks, R. and M. Harel (2006). "An economic game theory model of subcontractor resource
allocation behavior." Construction Management & Economics 24(8): 869-881.
Sacks, R., M. Radosavljevic, et al. (2010). "Requirements for Building Information Modeling based
Lean Production Management Systems for Construction." Automation in Construction in press.
Sacks, R., M. Treckmann, et al. (2009). "Visualization of Work Flow to Support Lean Construction."
Journal of Construction Engineering and Management 135(12): 1307-1315.
Seppanen, O. (2009). Empirical Research on the Success of Production Control in Building
Construction Projects. Faculty of Engineering and Architecture. Espoo, Finland, Helsinki University
of Technology PhD: 325.
Sriprasert, E. and N. Dawood (2003). "Multi-Constraint Information Management and Visualisation
for Collaborative Planning and Control in Construction." ITcon - IT in Construction 8: 341 - 366.
Synchro. (2007). "Synchro." from www.synchroltd.com.
Tekla. (2008). "Construction Management." from www.tekla.com.
VICO. (2007). "Virtual Construction 2007." Retrieved April 2007, from
http://www.vicosoftware.com/.
Womack, J. P. and D. T. Jones (2003). Lean Thinking: Banish Waste and Create Wealth in Your
Corporation. New York, Simon & Schuster.
13
http://www.synchroltd.com/http://www.tekla.com/http://www.vicosoftware.com/
-
RCM-Plan: A Computer Prototype for Improving Planning Reliability from a Lean Production Viewpoint
González, V.
Universidad de Valparaíso
(email: [email protected])
Alarcón, L.F.
Pontificia Universidad Católica de Chile
(email: [email protected])
Ulloa, H.
Pontificia Universidad Católica de Chile
(email: [email protected])
Abstract
The creation of reliable work plans level is perhaps one of the most relevant stages in construction
planning process. However, current practice and research in construction are characterized by
developing this process in an informal fashion in which decisions are based mainly on intuition and
experience of project personnel. This paper introduces RCM-Plan, a computer prototype designed to
support operational planning using the so-called Reliable Commitment Model (RCM). RCM is a
statistical modelling approach based on lean production principles, which produces reliable
predictions of work plans, capacity and other production variables for short term-periods using
common field information such as workers, buffers, and planned progress. Thus, RCM promotes a
more reliable operational planning process and improved performance of work plans. The core
functions of RCM-Plan are designed to systematize and automate the procedures associated with the
RCM methodology, generating the analysis and information in a quick, easy way and it is
instrumental to facilitate the application of the RCM concepts. The capabilities of RCM-Plan are
demonstrated using a case study. In addition, limitations and further research related to the RCM-
Plan are addressed.
Keywords: computer prototype, lean production, operational planning reliability, RCM-plan, reliable
commitment model
14
-
1. Introduction
The creation of reliable work plans has received increased attention as a critical issue in construction
projects performance during the last decades (Ballard, 2000; Tommelein et al, 1999; among others).
The recognition that these plans at the operational level characterize the materialization of a project,
namely its production through construction processes, highlights their decisive role for
accomplishment of project goals. In practice, a reliable work plan is characterized by a high
fulfillment of planned work. However, the decision frame to make and predict work plans has
several limitations in traditional management practices (González et al, 2009). Commonly,
construction projects outsource most of the work to subcontractors, and work plans are arranged
between contractors and subcontractors. Contractors should strive to obtain reliable work plans from
the subcontractors. Thus, many of them assign work to subcontractors based on their intuition and
experience, resulting in unreliable work plans (Sacks and Harel, 2006). Then, the more unreliable
work plans the lower is project performance (Ballard, 2000).
Lean production is a management philosophy focused on adding value from raw materials to finished
product. It allows avoiding, eliminating and/or decreasing waste such as waiting/idles times, rework,
overproduction, excessive movement, among others, from the value stream. One of the main goals of
lean production is reducing variability (Womack and Jones 1996). In construction, variability depicts
varying production rates, labor productivity, schedule control, cost control, etc., which is a well-
known problem due to its detrimental impacts in project performance on which there is much ongoing
research (Ballard, 1993; Tommelein et al, 1999; among others). The Last Planner System (LPS™) is
a production planning and control system based on lean production principles that was developed to
improve planning reliability in construction projects (Ballard, 2000). LPSTM
provides the means and
tools to deal with variability in projects providing a stable production environment and reducing the
negative impacts of variability. This helps create reliable work plans for short-term periods
(operational level). However, last planners still create and predict work plans to make their planning
commitments using mainly their intuition and experience, resulting sometimes in unreliable
commitments (González et al, 2009).
Thus, this paper proposes a computer prototype termed as RCM-Plan to help planners to make more
reliable performance predictions based on data obtained on site. RCM-Plan is based on the Reliable
Commitment Model (RCM) (Gonzalez et al 2010), a lean tool which enhances the creation process of
work plans in repetitive projects. RCM uses statistical models to predict performance and produce
more reliable work plans, using information about workers, buffers, actual and planned progress
avoiding current decision patterns. The core functions of RCM-Plan are designed to systematize and
automate the RCM procedures, generating analyses and information in a quick and easy way. In this
paper, planning reliability improvement using the LPS™, the RCM conceptual and analytical
framework and the computer structure of the RCM-Plan are addressed. Then, a case study is used to
show the use of the RCM-Plan to support the development of reliable work plans.
15
-
2. Improving planning reliability
The Last Planner System (LPS) is a production planning tool based on lean principles widely used
today in construction. LPS™ acts over four project planning levels: (i) „Initial planning or master
plan‟ (strategic level), which produces the initial project budget and schedule, and provides a co-
ordinating map that „pushes‟ completions and deliveries onto the project. (ii) „Phase Planning‟ a
subdivision of the master plan which is transformed in a “pull plan,” with participation of key
members of the project team during that phase. (iii) „Look ahead planning‟ (breakout of master plan
– tactical level), which details and adjusts budgets and schedules „pulling‟ resources into play. (iv)
„Commitment planning or work plans‟ (short-term period – operational level), which regards the
activities and schedule work that will be done on-site according to the status of resources and
prerequisites (Ballard, 2000).
The traditional management approach for work plans defines activities and schedule work that will be
done, in terms of what should be done from a master plan, compromising crews with no real
consideration for what they are actually able to do. Then, the crew ability to reliably perform work
depends on the stability of the so-called workflow. In construction, workflow can be characterized
by crews moving from location to location and completing the work that is prerequisite to starting
work by the following crew. In turn, a stable workflow depends on what construction preconditions
such as resource (design, components and materials, workers, equipment, space) and prerequisites
(complete work of upstream activities) should be available whenever they are needed. However,
variability of workflow could negatively affect crews‟ performance, causing idle time or ineffective
work (Tommelein et al, 1999).
In contrast, LPS™ provides a predictable production environment in projects, decreasing workflow
variability and creating reliable work plans to derive maximum project benefits. The overarching
criterion in the LPS™ is that activities should only be committed if they can be performed (i.e. all
resources and prerequisites that are needed must be available), transforming what should be done into
what can be done, from which a work plan can be formulated. Thus, work plans will be based on
achievable assignments serving as a commitment to what will actually be done. In this paper, the
notion of reliability is focused on project planning, so that Percentage of Plan Completed (PPC)
depicts a project planning reliability index. PPC is understood as the ratio between actual completed
activities and planned activities in a short term period (typically one workweek). A low PPC means
unreliable planning and a high PPC close to 100%, means the opposite.
LPS™ has been applied in numerous projects around the world in the last fifteen years, and a wide
range of performance improvements have been reported (Alarcon et al 2005, Liu and Ballard 2008,
González et al, 2008a). The main system assumption is that an increase in planning reliability,
measured through PPC, should improve project performance. Recently, several researchers have
demonstrated a positive and strong relationship between planning reliability and project performance
(González et al, 2008a). However, current planning practices at the operational level reduce the
ability to achieve reliable commitments to improve project performance. The RCM model is
designed to support the planners in making more reliable commitments in the short term planning
16
-
process and it can be used within the LPS process or as an individual method when the LPS is not in
place.
3. Reliable Commitment Model (RCM) framework
RCM provides an operational decision-making tool for predicting work progress in projects using
statistical models. These models use historical information of several production variables such as
labor, Bf, and planned progress, to attain a more reliable planning process at the operational level.
Since RCM is based on lean principles, it helps to reduce production variability by improving
planning reliability and matching work load with labor capacity (more details in González, 2010). In
practice, RCM uses multiple linear regression (MLR) to formulate the model, which assumes the
following form: (González, 2010):
PPWIPBfW PRP 3210 (1)
where:
PRP is the predicted progress for a process in a short-term period (typically one workweek). Units
may be m2, m
3, linear-meters, houses, apartments, etc.
W is the number of workers for a process in a short-term period. W is the sum of workers in the
planning horizon. For instance, if the short-term period is one workweek of 5 days, and there are 5
worker-days, W is 25 worker-weeks.
WIP Bf is the available Work-In-Process Buffer for a process at the beginning of a short-term period.
In general terms, a Buffer allows isolating a production process from the environment as well as the
processes depending on it. Buffers can prevent the loss of throughput, wasted capacity, inflated cycle
times, larger inventory levels, long lead times, and poor customer service by shielding a production
system against variability (more details in Hopp and Spearman, 2000). For instance, for a one
workweek the WIP Bf for the painting process, which depends on the wall-stucco process, is the
available work produced by the wall-stucco process, measured at the beginning of the week, before
painting begins. Units may be m2, m
3, linear-meters, houses, apartments, etc.
PP is the planned progress for a process in a short-term period. Units may be m2, m
3, linear-meters,
houses, apartments, etc.
Only significant variables are selected in the RCM models, since including redundant variables may
lead to incorrect analysis of scenarios. Thus, MLR models with the least number of variables and
with the highest coefficient of determination (R2) are selected. Bustamante (2007) demonstrated that
good quality models are obtained using this heuristic. Furthermore, the prediction accuracy of RCM
is evaluated using two indicators: Process Reliability Index (PRI) and Predicted/Planned
Commitment Confident Level (CCL). PRI measures the degree of process effectiveness from a
commitment standpoint, expressed as:
(2) 100PP
AP PRI
ji,
ji,
17
-
where:
PRIi,j is the process reliability index for week i and process j (%), i=1…n; j=1...m.
APi,j is the actual progress for week i and process j, i=1…n; j=1...m.
PPi,j is the planned progress for week i and process j, i=1…n; j=1...m.
PRI ranges between 0 and 100%. Note, when AP is higher than PP, PRI is limited to 100%. A low
PRI means unreliable process planning and a PRI close to 100% means the opposite. Meanwhile
Predicted/Planned CCL is a measure of the planning prediction reliability for a process made by both
decision-makers (planned progress) and/or RCM models (predicted progress) and it is expressed as:
(3)
where:
Predicted/Planned CCLi,j is the commitment confidence level for week i and process j (%) for both
predicted and planned PRI.
Predicted PRIi,j is the predicted process reliability index for week i and process j. Predicted PRI
replaces AP by PRP in Eq. (2).
Planned PRIi,j is the planned process reliability index for week i and process j. Its value is estimated
by a decision-maker given a planned progress according to his or her experience.
Actual PRIi,j is the actual or real process reliability index for week i and process j. Actual PRI is
computed using Eq. (2).
Predicted CCL measures the process commitment accuracy for the predicted progress comparing the
predicted and actual PRI. Similarly, Planned CCL compares the planned and actual PRI. When the
ratio in Eq. (3) is less than 0, its value is set to 0.
RCM originally was implemented through nomographs, which relate mathematical and graphically
planned progress with the other production variables (González et al, 2010). Then, since PRI is given
by:
PPPRIPRPPP
PRPPRI (4)
And, if equation (1) is replaced in equation (4), PP can be expressed as:
3
210
PRI
WIPBfWPP (5)
Eq. (5) establishes a relationship between PP and W, WIPBf, and PRI. PP can be either planned by
decision makers or estimated by the RCM. Similarly, PRI can be either planned or predicted. Fig. 1
illustrates a nomograph for a repetitive housing project, showing the interaction between the different
variables involved. Nomographs are commonly used in engineering disciplines (e.g., hydrologic
engineering) and can be easily applied by construction decision-makers, such as project managers,
which can use it to plan activity progress for a given resource frame. RCM methodology to estimate
100ActualPRI
ActualPRIPlannedPRI/edictedPr1PlannedCCL/edictedPr
j,i
j,ij,i
j,i
18
-
work plans is summarized in Fig. 2. The conceptual and mathematical RCM framework, as well as
the implementation methodology, were comprehensively tested and validated in 9 projects
(multifamily residential, multi-story building and industrial), 23 activities, and a total of 260
workweeks by González et al (2010).
Figure 1: Illustration of a general RCM nomograph to calculate planned progress for next short-term
period, given a pre-determined MLR model (with the assumption that it is a reliable model such that
R2 ≥0.6 and P-value ≤ α=0.05) and different PRI values (adapted from González et al, 2009)
1. Selection of activities
2. Initial data collection for each activity in week i: PP, planned PRI, planned W
WIPBf, AP and actual W.
i) Is the measured horizon plan higher than two weeks?
3. Selection of the best MLR model using the number of variables, R2 and P-values
for week i+2
No
Yes
ii) Is it possible to determine a valid
regression model?No
Yes
4. Def inition of Nomographs for week i+2
5. Def inition of a base case for week i+2 given initial data in which a predicted PRI
value is def ined.
10.Evaluation of the RCM predictions accuracy for week i+2. Indicators:
i)Predicted and Actual PRI, and ii) Predicted
Start
v) Does the RCM process
continue?Finish
No
Yes
Note: To construct a multiple linear regression model is necessary at least two data sets or points. Then, it is the reason to request two planning weeks measured in the f irst decision block.
9.Final data collection for each activity in week i+2: AP and actual Wiii) Is predicted
PRI
-
sections describe previous tools for improving planning reliability and production predictions in
construction. Then, the RCM-Plan software architecture is introduced.
4.1 Computer approaches to improve planning reliability and production predictions in construction
LPS™ application has inspired the development of several computers applications, supporting
several of its main functions, which allow improving planning reliability. Among others, can be
mentioned: WorkPlan (Choo et al, 1998); WorkMovePlan (Choo and Tommelein, 2000); Integrated
Production Scheduler (IPS) (Chua et al, 1999); Lean Enterprise Web-based System for Construction
(LEWIS) (Sriprasert and Dawood, 2002); SPS Production Manager (SPS|PM) (SPS|PM, 2009);
Project Plus Control (P+C) (P+C, 2009). In general, these approaches provide support to the
planning process but do not solve the operational planning issue that emerges when contractors and
subcontractors arrange work plans making predictions about their performance. In general, many
contractors assign work to subcontractors based on their intuition and experience resulting in
unreliable work plans.
On the other hand, there are several methods in construction that can improve production predictions
such as virtual prototyping (Huang et al, 2007); or discrete event simulation modeling (Martínez,
1996). However, these predictive approaches can be complex to adopt among construction
practitioners as an ordinary practice. RCM-Plan, described in the next section, provides a tool that is
easy to use and implement to automate the procedures associated with RCM, providing prediction
capabilities to support the adoption of reliable commitments in planning meetings.
4.2 RCM-plan framework
RCM-Plan was developed as part of a research effort from the Production Management Center
(GEPUC) at Pontificia Universidad Católica de Chile, to improve the current planning practices in
the Chilean construction industry. RCM-Plan architecture is based on Microsoft Visual Basic .Net,
framework 2.0, developed with Visual Basic 2005. Microsoft Excel version 2000 is required for
RCM-Plan operation on the client-machine user application. The matrix approach is used in RCM-
Plan to calculate the different parameters of the MLR models, i.e. β0…β3, which is implemented in
Excel and in turn interacts with the .Net framework, showing the main MLR statistical results. By
using Excel the system generates the necessary graphics such as nomographs, weekly progress, PRI
evolution, labor resource evolution, etc. and several types of reports. RCM-Plan does not use data
base provided by external suppliers, since its arquitecture uses .xml files to store historical data,
making the information very portable. By summarizing RCM methodology shown in Fig. 2, RCM-
Plan computer architecture and operation is briefly described as follows:
1) Selection /Creation of Activities: In this stage, project managers select activities to improve their
planning reliability. Thus, activities are created within RCM-Plan and are individually analyzed to
20
-
get their different production outputs and statistical parameters. For one or more activities, RCM-
Plan generates data base files termed as „.rcmpl‟ that can be imported or exported to other computers.
2) MLR Models and Nomograph Construction: The RCM process collects on-site information
and predicts work plans performance on a weekly basis, generating typically MLR models starting
from the 3rd
week. Figure 3 shows a general screen for tools and information that provides RCM-
Plan, listing a specific activity (in this case, partition-joint tape). The main tools are related to Enter
Planning (for actual week), Modify Planning of Last Week, Modify Selected Week (data of any week),
PRI and Progress Graphics (PRI and Progress evolution) and Average PRI (summary of PRIs and
CCLs). In this version, only the predicted CCL is regarded. Also, RCM-Plan considers a predicted
PRI equal to 100% to develop nomographs (a common reliability level expected by planners). In
Figure 3, 17 weeks of historical information are considered and the planning process of 18th week is
illustrated.
Partition Joint-Tape (ML)
Workers (W)
Workers (W)
Workers (W)
Figure 3: General screen of RCM-Plan
To plan 18th week, it is necessary to select a MLR model. RCM model automatically calculates all the
possible combination of variables (W, WIPBf and PP) and its corresponding R2. In Figure 4, the
Adding Planning for Week 18 dialog box, Selection of Function tab, shows which variables are
selected for a specific MLR model and the corresponding R2-value (decision-makers manually chose
a MLR model using the heuristic mentioned earlier in section 3). This dialog box allows showing the
full MLR model (View Prediction Function option) which is a function depending on W and WIPBf
variables in this case and creating the corresponding nomographs (Nomograph option). The latter
option allows showing graphically the interactions between PP and W given several WIPBf sizes.
This is the most important feature of the RCM-Plan, to explicitly visualize in a multidimensional
environment the interactions of several production parameters, allowing reliable prediction of activity
progress, and thus, more reliable planning commitments and work plans.
3) RCM Planning Process: Once a decision-maker has selected PP, W and WIPBf levels, this
information should be entered into the planning for next week (18th week in Figure 5), which is done
in the Adding Planning Week 18 dialog box, Enter Planning Data tab. Note that W is entered as a
planned estimate. Predict Progress is automatically generated as RCM output, showing the PRP level
21
-
for week 18, i.e. predicted work plans for this week. At the end of the 18th week, AP and actual W
levels are entered using the Modifying Progress dialog box.
Figure 4: Adding planning by creating RCM prediction function and nomographs
Figure 5: Adding planning by entering RCM inputs and actual progress
4) Evaluation and Feedback: At the end of every week, several reports are produced (see Figure 6).
5000 (ML)
3000 (ML)
8000 (ML)
2800
2200
Redefine Chart Limits Print Chart
Worker-Weeks (W)
Addi ng Planning Week 18
Pla
nned
Pro
gres
s (P
P)
MLWorker-Weeks
Actual Progress
Enter progresses and Worker-
Weeks, corresponding to the week 18
Planned progress for this week: 2500
(ML), with 20 planned Worker-Weeks
22
-
Figure 6: PRI and progress evolution reports
The PRI and Progress Graphics dialog box describes the Predicted/Actual PRI and
Planned/Predicted/Actual Progress respectively for all weeks. Also the Summary of PRI and
Predicted CCL dialog box described the average Predicted/Actual PRI and the average Predicted
CCL for the analysis period. This information allows evaluating performance of planning process in
terms of its reliability and accuracy using the RCM-Plan.
5. Case study
From the total cases studied with the RCM (section 3), RCM-Plan was actively implemented in 6
projects and 15 activities. The RCM-Plan arquitecture allows the automation of RCM calculations
and procedures and a reliable visualization of its mains inputs. These features facilitate the on-site
application of RCM concepts and reduce possible barriers of implementation. In this section, a case
study where RCM-Plan was applied is used to illustrate its impacts on planning reliability and
activity performance.
5.1 Case study A: Multi-family residential building
Plastering activity in a multi-family residential building was selected, analyzing how improving
planning reliability through the RCM-Plan could increase labor productivity. Figure 7 summarizes
the data evolution of RCM-Plan application.
PRI and Progresses Graphics
Select number of weeks to view Print Graphics
Week
Week
Actual PRIPredicted PRI
Planned ProgresPredicted Progres
Actual ProgresWe
ek
%P
lan
nin
g R
elia
bili
ty
23
-
No-predictions period
0
200
400
600
800
1000
1200
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17
Pro
gre
ss
(lm
)
Week Planned ProgressPredicted ProgressActual Progress
+∆ PRI 41.0% +∆ productivity 10.3%
Predictions/no-decisions period
Predictions/decisions period
Mean Labor Productivity 20.4 (m2/worker-weeks)
Mean Labor Productivity 22.5 (m2/worker-weeks)
No-predictions period
Figure 7: RCM application evolution case study A (González et al, 2009)
In this case, MLR models were mainly specified as a combination of W and/or WIPBf variables, with
different model parameters, β1 and β2, every week (note that RCM process is dynamic, changing
parameters and even variables of MLR models week to week). Three different periods can be
distinguished in Figure 7: the No-predictions period (1st
– 2nd
weeks), where data is collected for the
RCM; the Predictions/no-decisions period (3rd
– 13th weeks), where planning predictions were
performed to test the RCM, but were not used by the manager to make decisions; and the
Predictions/decisions period (14th -17
th weeks), where the manager relied on the RCM outputs to
make planning decisions.
RCM-Plan was used in the planning process of this case from the beginning, being the RCM-Plan
impact analysis focuses on the last two periods. An active intervention on the W and WIPBf
variables was decided on the 11th week, starting from the 14
th week, when sensitivity analyses using
the RCM-Plan were performed to study the effect of WIPBf over W. Based on these analyses, the
manager decided to slow down the activity not involving a higher number of W during 14th and 15
th
weeks. During these weeks, a larger WIPBf was deliberately generated maintaining a low W level. It
was determined that a WIPBf size closer to 2000 m2 could maximize labor productivity in order to
achieve PP levels of 800 m2 with W levels closer to 31 worker-weeks. During the 16th and 17
th
weeks the numbers of W was increased to take advantage of a higher WIPBf size.
A rough analysis of the data from Figure 7 shows that the mean actual PRI for the Predictions/no-
decisions period and the Predictions/decisions period is 70.55% and 100%, respectively. The effect
over labor productivity for the same period was estimated as the ratio between actual progress and
worker-weeks. Mean labor productivity for the Predictions/no-decisions period and
Predictions/decisions period was 20.4 (m2/worker-weeks) and 22.5 (m
2/worker-week), respectively.
In other words, planning reliability was increased by 41.0% and productivity by 10.3% (see Figure
7). A larger WIPBf size during weeks 14th and 15
th was produced, resulting in improved productivity
for the following weeks. The mean actual PRI for the 3rd
- 15th weeks is 72.49% and for the 16
th - 17
th
weeks is 100%, improving planning reliability by 38.0%. Similarly, the mean labor productivity for
24
-
the 3rd
- 15th weeks is 20.6 (m
2/worker-weeks) and for the 16
th - 17
th weeks is 27.0 (m
2/worker-weeks),
with a productivity improvement of 31.0%. The improvement in labor productivity is explained by a
better planning reliability and a direct action over production variables such as W and WIPBf using
the RCM-Plan as a decision-making aid tool.
6. Conclusions
This paper addressed the theoretical and practical framework that supports a computer prototype
called RCM-Plan based on Rational Commitment Model (RCM) principles, which allows to improve
reliability of operational work plans in construction projects. By using site production data such as
workers, buffers and plans, RCM-Plan automatically develops statistical models (multiple linear
regressions) to predict activity progress, and thus, support reliable work plans. One of the most
important features of the RCM-Plan is the explicit visualization in a multidimensional environment of
the interaction of several production parameters (workers, buffers and plans) and its impact on
planning predictions and labor capacity. This allows creating more reliable work plans, having in
mind a more realistic and a rational characterization of the production environment in projects. The
implementation of the RCM methodology can be complex but experimentation with RCM-Plan
allowed reducing implementation barriers in the organization of the case projects. Further research to
improve capabilities of RCM-Plan is necessary, which is part of the ongoing research of the authors.
References
Alarcón, L.F., Diethelm, S., Rojo, O. and Calderón R. (2005). Assessing the Impacts of Implementing
Lean Construction. Proceedings 13th Annual Conference of International Group for Lean
Construction, Sidney, Australia, July 19th – 21th.
Ballard, G. (2000). The Last Planner System of Production Control. PhD’s Dissertation, School of
Civil Engineering, Faculty of Engineering, The University of Birmingham, Birmingham, U.K.
Bustamante, J. A. (2007). Implementing of a Rational Commitment Model for Improving Planning
Reliability in Construction Projects. MEng’s Thesis, Faculty of Engineering, Pontificia Universidad
Católica de Chile, Santiago, Chile.
Choo, H.J. and Tommelein, I.D. (2000). WorkMovePlan: Database For Distributed Planning And
Coordination. Proceedings 8th Annual Conference of International Group for Lean Construction,
Brighton, UK, July 17th-19
th.
Choo, H.J., Tommelein, I.D., Ballard, G. and Zabelle, T.R. (1998). WorkPlan: Database for Work
Package Production Scheduling. Proceedings 6th Annual Conference of International Group for Lean
Construction, Guaruja, Brazil, August 13th-15
th.
25
-
Chua, D.K.H., Shen, L.J. and Bok, S.H. (1999). Integrated Production Scheduler for Construction
Look-Ahead Planning. Proceedings 7th Annual Conference of International Group for Lean
Construction, Berkeley, USA, July 26th-28
th.
González, V., Alarcón, L.F., Maturana, S., Mundaca F. and Bustamante, J. A. (2010). Improving
Planning Reliability and Project Performance using the Reliable Commitment Model. Journal of
Construction Engineering and Management, 10.1061/(ASCE)CO.1943-7862.0000215 (in-press).
González, V., Alarcón, L.F., Maturana, S., Mundaca, F., and Bustamante, J. A. (2009). Rational
Commitment Model: Improving Planning Reliability and Project Performance. Proceedings 17th
International Conference for Lean Construction, Taipei, Taiwan, July 15th – 17
th.
González, V., Alarcón, L.F. and Mundaca, F. (2008a). Investigating the Relationship between
Planning Reliability and Project Performance. Prod. Plan. Control, 19, (5), 461-474.
González, V., Alarcón. L.F., Maturana, S., Bustamante, J.A. and Mundaca, F. (2008b). Work-In-
Process Buffer Management using the Rational Commitment Model in Repetitive Projects.
Proceedings 16th Annual Conference of International Group for Lean Construction, Manchester,
UK, July 16th-18
th.
Huang, T., Kong, C.W., Guo, H.L., Baldwin, A. and Li, H. (2007). A Virtual Prototyping System for
Simulating Construction Processes. Autom. Constr., 16, (5), 576-585.
Hopp, W. J. and Spearman, M. L. (2000). Factory Physics: Foundations of Manufacturing
Management, 2nd
Ed., Irwin/McGraw-Hill, Boston.
Martínez, J. C. (1996). STROBOSCOPE: State and Resource Based Simulation of Construction
Processes. PhD’s Dissertation, University of Michigan, Ann Arbor, Mich.
P+C (2009). Project Plus Control. Accessed on December 12th, 2009.
See www.projectpluscontrol.cl/noticias.html
Sacks, R. and Harel, M. (2006). An Economic Game Theory Model of Subcontractor Resource
Allocation Behaviour. Const. Mgmt. and Econ., 24, (8), 869–881.
SPS|PM (2009). SPS|Production Manager. Accessed on December 12th, 2009.
See www.strategicprojectsolutions.com/Products_Services_SPSPM.aspx.
Sriprasert, E., Dawood, N. (2002). Next Generation of Construction Planning and Control System:
The Lewis Approach. Proceedings 16th Annual Conference of International Group for Lean
Construction, Gramado, Brazil, August 6th-8
th.
Tommelein, I. D., Riley, D. R. and Howell G. A. (1999). Parade Game: Impact of Work Flow
Variability on Trade Performance. J. Const. Engr. Mgmt., ASCE, 125, (5), 304-310.
26
http://www.projectpluscontrol.cl/noticias.htmlhttp://www.strategicprojectsolutions.com/Products_Services_SPSPM.aspx
-
Womack, J. P. and Jones, D. T. (1996). Lean Thinking: Banish Waste and Create Wealth in Your
Corporation, Simon & Schuster, New York.
27
-
Building Information Modelling in the Netherlands: A Status Report
van Nederveen, S.
Delft University of Technology
(email: [email protected])
Beheshti, R.
Delft University of Technology
(email: [email protected])
Willems, P.
TNO
(email: [email protected])
Abstract
Building Information Modelling (BIM) is nowadays widely accepted as a key enabler for innovation
in construction. In the Netherlands, people have been working on BIM for more than twenty years,
although most activities have been research efforts. But since the leading CAD vendors have
embraced BIM as a key development in CAD innovation, the implementation and use of BIM
technologies in practice have increased significantly.
Apart from various “pseudo-BIM”- initiatives (BIM-solutions within a single commercial software
platform, “closed” BIM-solutions that are not accessible by external parties, fancy CAD-solutions
presented as BIM-solutions), there are a number of interesting BIM-related developments in the
Netherlands.
The first development is the COINS project. This project aims for agreements for the storage and
exchange of construction objects. The main results of COINS are currently specifications for these
agreements and software tools for implementation of COINS-based systems. COINS uses the OWL
format for object definitions; interfaces with the IFC-models have also been developed. The COINS
project is initiated by the Dutch civil engineering industry, but the current focus is on products of the
entire building industry. Several pilot projects are currently taking place both in civil engineering and
in office and residential building.
The second development is somewhat different: it is the BIM Case Week, an initiative that brings
together professionals in the construction industry for a week, and lets them work together on a
design of a building project. The approach is fairly down-to-earth but it has provided very useful
insights in how exchange and sharing of construction information takes place in practice.
28
mailto:[email protected]
-
The third development is the Dynamic BIM initiative; this is currently an academic initiative that
aims at the support of project dynamics in a BIM context. The focus in this initiative is on innovative
design and engineering processes enabled by BIM technologies.
Keywords: building information modelling, the Netherlands, status report
29
-
1. Introduction
One of the current keywords in building innovation is Building Information Modelling, or BIM.
Since a few years, BIM is a buzzword. But Building Information Modelling is not a new activity or
technology; in fact people have been working on BIM (using different terms) for decades, although
mainly in research settings.
People in the Netherlands have been involved in BIM research since the early days of BIM, the
eighties of last century. Over the years, Dutch researchers have been working on BIM in various
projects. In recent years, a transition can be seen in the Netherlands from mainly research oriented
activities towards dissemination and implementation in building practice. More and more people and
companies become active with BIM. Articles on BIM appear in practice oriented journals, new
courses for building professionals are set up, and new dedicated BIM websites are set up, and so on.
This paper gives an overview of some important BIM-related developments in the Netherlands. First
a short historical overview of BIM in the Netherlands is given. Next the COINS project, the BIM
Case week and the Dynamic BIM initiative are discussed, followed by a short discussion of other
developments. But before all that, a short statement is made about the definition of BIM.
2. What is BIM?
There are a number of different definitions of BIM around. As more people are working with BIM,
this number increases, and as a consequence more misunderstandings occur. As long as BIM is
mainly a research topic, this is a little unpractical, but more or less unavoidable, just like with many
other definitions in research. But with the transition from BIM as a research topic towards BIM as a
commercial product or service, the need for a clear definition becomes really apparent. For example,
many companies claim they are doing BIM while critics say they are only offering smart CAD
solutions.
A useful definition for the term Building Information Modelling is the following by Lee et al (2006),
which is also used on Wikipedia: Building Information Modeling (BIM) is the process of
generating and managing building data during its life cycle. Typically it uses three-dimensional,
real-time, dynamic building modeling software to increase productivity in building design and
construction. The process produces the Building Information Model (also abbreviated BIM), which
encompasses building geometry, spatial relationships, geographic information, and quantities and
properties of building components
In addition, a useful definition for the term Building Information Model is the following by Van
Nederveen et al (2009): a Building Information Model is an information model of a building (or
building project) that comprises complete and sufficient information to support all lifecycle
processes, and which can be interpreted directly by computer applications. It comprises information
about the building itself as well as its components, and comprises information about properties such
as function, shape, material and processes for the building life cycle.
30
http://en.wikipedia.org/wiki/Building
-
The last definition is a little bit long term oriented, as life-cycle support by BIM is currently far from
common practice.
But let us go back to the initial question: what is BIM? A key question in this respect is: how can we
distinguish between “BIM” and “non-BIM”? For that purpose, the following characteristics of BIM
can be highlighted:
BIM aims at the exchange of semantic information. That is: the model that is developed does not only cover geometric information, but also material properties, functional information,
etc. For example, many advanced CAD systems that use concept of parametric modelling can
be very useful design aids. But if their internal model is solely based on geometric entities,
you cannot call these BIM modellers.
A prerequisite for BIM is the use of open standards. A Building Information Modeller may be a “closed” system, but the information that is exchanged or shared must be defined
according to an open standard, such as IFC. Although closed systems can be very effective, in
the long run they can lead to vendor-dependency and to outdated systems that are very
difficult to upgrade.
Neither of the definitions stated above explicitly mention open standards as a prerequisite for proper
BIM. Open standards are indeed often mentioned as a prerequisite. On the other hand, one can
question whether it is absolutely necessary to use for example IFCs in a BIM environment. In our
view this is an open issue.
3. History of BIM in the Netherlands
The Netherlands has quite a rich history in BIM research and development, which goes back to more
than twenty years ago. In the nineteen eighties, several groups in the Netherlands were involved in
research on CAD systems for architecture, and on the issue of data exchange between CAD systems.
The Dutch architectural CAD system Arcos/Arkey CAD was launched with some “building
intelligence” built in. A discussion started on the use of so-called reference models for CAD
exchange.
A key reference model in this context was the General AEC Reference Model by Wim Gielingh of
the Dutch research institute TNO (1988). This model was developed for the ISO STEP (ISO 10303)
project, and it provided a number of concepts and principles that we can regard now as BIM
concepts: as required and as designed information, generic-specific-occurrence information, life cycle
data, views on building data, etc. The famous “Hamburger” notation and the associated ideas are still
used in publications from all over the world.
Another interesting publication out of that period is the so-called IOP Bouw Informatie Model (Van
Merendonk and Van Dissel 1989). This model was the main end result of a large Dutch research
project aiming at the modelling of building information. Most of this publication consists of process
31
-
models in IDEF0, furthermore some data models have been presented in IDEFx. The models are
nowadays rarely used or referenced, but the title of the model is definitely remarkable.
In the early nineties, some very interesting BIM-related work was carried out in EU-projects in which
TNO was involved, such as ATLAS, PISA and COMBINE (Tolman 1999). In all of these projects,
product modelling based on ISO STEP played a key role. Many key concepts and principles of IFC
origin from these projects.
From the late nineties until today, a number of smaller scale national activities related to BIM took
place. Participants involved include among others the building specification organization STABU,
the organization for installation systems UNETO. Some of the initiatives have formed a platform
called PAIS, see www.paisbouw.nl. A significant national development has been VISI, a standard for
communication in building projects based on transactions and messages, see www.visi.nl. VISI uses
protocols for common communication processes using transactions that consist of a sequence of
messages between participants.
At the moment there are a number of interesting BIM developments going on in the Netherlands.
Three developments will be discussed in the next three sections of this paper. The first development
is the COINS project. This project is interesting because many key players from the Dutch
construction industry are participating. The approach taken can be regarded as pragmatic, yet they do
use an open standards approach based on IFC and OWL.
The second development is the BIM Case Week. This initiative brings together professionals in the
construction industry for a week, and lets them work together on a design of a building project. The
BIM Case Week is similar to the Build London Live events in the UK. Its biggest value is the great
amount of public attention for BIM that it attracts.
The third development is the Dynamic BIM initiative. This is currently an academic initiative that
aims at the support of project dynamics in a BIM context. This initiative is particularly interesting
because it tries to bring BIM another step further through new research and innovation.
An important development in the Netherlands that is not directly about BIM, but that has a significant
impact on BIM work, has been the growing interest in Systems Engineering. Since the late nineties,
Systems Engineering was introduced at the large infrastructure principals ProRail and
Rijkswaterstaat, when these organizations became involved in large scale projects such as the High
Speed Link railway project between Amsterdam and Paris. With Systems Engineering, infrastructure
projects became more formal, with explicit procedures for requirements management, verification &
validation and risk management. Of course the companies that work for Rijkswaterstaat and ProRail
had to follow the Systems Engineering process, which meant that almost the entire civil engineering
sector had to deal with Systems Engineering. The impact of Systems Engineering on BIM work in the
Netherlands can be seen in current developments that will be discussed below.
32
http://www.paisbouw.nl/http://www.visi.nl/
-
4. The COINS project
The first Dutch development to be discussed in this paper is the COINS project. This project aims for
agreements for the storage and exchange of construction objects. The acronym COINS stands for
„Construction Objects and the INtegration of processes and Systems (see www.coinsweb.nl and click
on “Introduction COINS program”).
The COINS project was started in 2003 by a number of organizations from